Abstract:Let{Xn} be a sequence of semimartingale in a filtered complete probability space (Ω,F,F,P) satisfying the usual condition. We use the general Girsanov thorem and closed graph theorem to prove that the sequence {Xn}converges on X in the Emery topology w.r.t Q if{Xn} converges on X in the Emery topology w.r.t P and the probability measureQloc《P. In light of this fact, we prove that if X is a d-dimensional semimartingale and a d-dimensional predictable process,H is X-integrable in the sense of vector stochastic integrals w.r.t P, when the probability measure Qloc《P, H is also X-integrable in the sense of vector stochastic integrals w.r.t Qand these two integrals are Q-differentiable. It is noted that the condition of Q《P is stronger than that of Qloc《P, therefore, this paper generalizes lemma 4.9 and theorem 4.14 in[1].